Jailbreak Defense on DrAttack
0ASRSelfGrader
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SelfGraderQ=10, Latency (Sec.)=0.74, Memory Overhead (MB)=2765.652026.04 | 0 | 0 | |
| SelfDefendBackbone=Qwen2.5-7B-Instruct, Latency (Sec.)=2.02, Memory Overhead (MB)=13137.25, Mode=Intent2026.04 | 0 | 13 | |
| SelfGraderBackbone=Qwen2.5-7B-Instruct, Latency (Sec.)=1.33, Memory Overhead (MB)=374.252026.04 | 0 | 0 | |
| GuardReasonerBackbone=Qwen2.5-7B-Instruct, Evaluation Stage=Post2026.04 | 1 | 75 | |
| WildGuardBackbone=Qwen2.5-7B-Instruct, Latency (Sec.)=0.94, Memory Overhead (MB)=13995.25, Evaluation Stage=Post2026.04 | 5 | 80 | |
| GuardReasonerBackbone=Qwen2.5-7B-Instruct, Evaluation Stage=Pre2026.04 | 8 | 36 | |
| GradientCuffLatency (Sec.)=40.58, Memory Overhead (MB)=2818.342026.04 | 9 | 13 | |
| SelfDefendDefense Protocol=Intent, Latency (Sec.)=2.14, Memory Overhead (MB)=13194.112026.04 | 10 | 15 | |
| WildGuardBackbone=Qwen2.5-7B-Instruct, Latency (Sec.)=1.00, Memory Overhead (MB)=13939.25, Evaluation Stage=Pre2026.04 | 10 | 50 | |
| SelfDefendBackbone=Qwen2.5-7B-Instruct, Latency (Sec.)=0.64, Memory Overhead (MB)=13130.75, Mode=Direct2026.04 | 11 | 58 | |
| Llama GuardBackbone=Qwen2.5-7B-Instruct, Latency (Sec.)=0.45, Memory Overhead (MB)=13739.13, Evaluation Stage=Post2026.04 | 15 | 95 | |
| GradientCuffBackbone=Qwen2.5-7B-Instruct, Latency (Sec.)=80.70, Memory Overhead (MB)=1672.502026.04 | 16 | 52 | |
| Qwen2.5-7B-Instruct (No Defense)Backbone=Qwen2.5-7B-Instruct, Latency (Sec.)=5.292026.04 | 19 | — | |
| Perplexity FilterBackbone=Qwen2.5-7B-Instruct, Latency (Sec.)=0.10, Memory Overhead (MB)=13134.882026.04 | 19 | 100 | |
| Prompt GuardBackbone=Qwen2.5-7B-Instruct, Latency (Sec.)=8.48, Memory Overhead (MB)=1154.582026.04 | 19 | 94 | |
| Llama GuardBackbone=Qwen2.5-7B-Instruct, Latency (Sec.)=0.36, Memory Overhead (MB)=13592.75, Evaluation Stage=Pre2026.04 | 19 | 84 | |
| SelfDefendDefense Protocol=Direct, Latency (Sec.)=0.84, Memory Overhead (MB)=13183.912026.04 | 36 | 48 | |
| Llama GuardDeployment Stage=Post, Latency (Sec.)=0.56, Memory Overhead (MB)=13753.872026.04 | 38 | 60 | |
| Llama GuardDeployment Stage=Pre, Latency (Sec.)=0.49, Memory Overhead (MB)=13649.822026.04 | 64 | 84 | |
| Prompt GuardLatency (Sec.)=15.11, Memory Overhead (MB)=1064.732026.04 | 77 | 99 | |
| Vicuna-13B-v1.5Defense=No Defense, Latency (Sec.)=3.202026.04 | 78 | — | |
| Perplexity FilterLatency (Sec.)=0.17, Memory Overhead (MB)=13214.012026.04 | 78 | 100 |